LT5903 Chap.8 How Children Acquire Language
How Children Acquire Language
Theories need distinguishing predictions
Language acquisition is explained through innateness, imitation, reinforcement, active construction, connectionist learning and social interaction. Overgeneralised forms matter because they can reveal productive rules or statistical generalisation rather than simple copying.
Research methods translate infant behaviour into evidence through sucking rate, head turns and preferential looking, while corpora record interaction over time.
Statistical word segmentation uses transitional probability, but evidence from natural input shows that stress and other cues may be necessary.
Compare explanations through novel behaviour
A child who produces a regularised form that adults do not use has gone beyond direct repetition. The form can support a productive rule, analogy or learned statistical pattern.
It does not by itself decide between innate-rule and connectionist accounts because both can predict generalisation under some conditions. Ask how each theory represents prior knowledge, input frequency, correction and exceptions, then derive a case where their predictions differ.
Evidence is most useful when it could have turned out otherwise.
Keep the developmental path visible
Early productions are not simply incomplete adult sentences. Children may show systematic stages, changing productivity and sensitivity to input. Track when a form first appears, whether it generalises to new items, how exceptions are treated and whether social interaction changes performance.
Cross-sectional comparisons show age-related differences, while longitudinal corpora reveal change within the same child. Neither design automatically controls input or individual variation, so match the conclusion to the sampling method.
Translate behaviour into an inference chain
Infant methods use an observable response as a proxy for attention or discrimination.
In high-amplitude sucking, a response can habituate and then recover after a new stimulus. Head-turn preference measures orientation or listening duration, and preferential looking compares gaze toward displays. For each method, identify the manipulated contrast, measured behaviour, baseline and inferential bridge.
A longer look is not itself knowledge of grammar; it becomes evidence only under a design that rules out simpler perceptual preferences.
Calculate transitional evidence carefully
Transitional probability estimates how strongly one unit predicts the next by dividing the frequency of a sequence by the frequency of its first unit in the relevant position.
High within-word values and lower boundary values can support segmentation in controlled streams. Preserve numerator, denominator and direction because the reverse transition may differ. Natural speech also contains stress, phonotactic, prosodic and contextual cues.
A successful laboratory calculation therefore demonstrates sensitivity to a cue, not that the cue is sufficient for every real input.
Use interaction as evidence, not a slogan
Social approaches emphasise joint attention, communicative intention, feedback and structured participation. To evaluate them, state which interaction differs and what change in learning is predicted.
Reinforcement likewise needs a specified response and consequence rather than a general claim that adults reward correctness. Imitation explains reproduced forms but struggles with novel errors unless combined with abstraction. Active construction focuses on the learner's evolving system. A balanced comparison explains each account's mechanism, supporting observation and unresolved limitation.
What this chapter covers
- 01
Innateness and critical periods
- 02
Imitation and reinforcement limits
- 03
Active construction and connectionism
- 04
Infant research paradigms
- 05
Transitional probability and segmentation
Worked example · free
Separate theories that fit one error
- 3State the active-construction account.
- 3State the connectionist account.
- 2Name distinguishing evidence.
Key terms
- Universal Grammar
- A proposed inborn set of structural characteristics shared by languages.
- Overgeneralisation
- Extension of a pattern to a form that is exceptional in adult language.
- Connectionist Model
- A learning model that adjusts weighted connections from input patterns.
- Head-Turn Preference
- A listening-time method used with infants.
- Transitional Probability
- The likelihood that one sequence element follows another.
How Children Acquire Language FAQ
Why is imitation insufficient as a complete theory?
Children produce novel sentences and systematic forms that adults do not provide, so direct copying cannot explain the whole pattern. Imitation may contribute input, but a learning account must explain productive generalisation beyond heard examples.
What does longer infant looking or listening prove?
It supports discrimination or preference under the experimental conditions. It does not directly prove an adult linguistic category or the mechanism causing the response. State the observable first and keep the inference appropriately narrow.
Why can transitional probability need help?
Artificial streams can make within-word and between-word transitions unusually clear. Natural speech contains varied sequences, and local syllable statistics may perform poorly without cues such as stress. A combined model can therefore outperform a probability-only account.
Exam move
Create a comparison grid with mechanism, supporting observation, weakness and decisive experiment for every theory. For methods, rehearse the observable-to-inference chain and one alternative explanation. Recalculate transitional probabilities in both directions and label the numerator and denominator.
For each infant paradigm, identify the baseline, manipulation, behavioural measure and control that would distinguish language learning from a simpler novelty or attention effect.
Working through How Children Acquire Language in LT5903? Sia is AskSia’s AI Arts and Humanities tutor — ask any LT5903 How Children Acquire Language question and get a clear, step-by-step explanation grounded in how LT5903 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
